Almost Timely News: 🗞️ Why You're Not Getting Enough Value from AI

(2024-12-08)

Dec 08, 2024

Almost Timely News: 🗞️ Why You're Not Getting Enough Value from AI (2024-12-08) :: View in Browser

The Big Plug

🚍 Take my new, completely free course: how to use Generative AI to land your next job! (and share it with anyone looking for work)

Content Authenticity Statement

99% of this week's newsletter was generated by me, the human. You'll see a few translations from Google Gemini in the opening section. Learn why this kind of disclosure is a good idea and might be required for anyone doing business in any capacity with the EU in the near future.

Watch This Newsletter On YouTube 📺

Click here for the video 📺 version of this newsletter on YouTube »

Click here for an MP3 audio 🎧 only version »

What's On My Mind: Why You're Not Getting Enough Value from AI

We're absolutely blowing it with AI right now.

Not because the tech isn't ready. Not because it doesn't work. We're blowing it because we're thinking way too small. We're taking this transformative technology and using it to write blog posts and social media updates - the equivalent of buying a Bugatti Chiron just to take the dog to the park.

This week, let's dig into why most AI implementations aren't delivering nearly the value they could be, what the skeptics are missing (and what they're getting right), and most important - what the path forward looks like if we want to extract real, meaningful value from AI.

Here's the thing: right now, almost every company I talk to is doing exactly what companies did in 1994 with websites - taking their existing processes, slapping some AI on top, and wondering why they're not seeing amazing results. We did the same thing back then, turning paper brochures into digital brochures and calling it "digital transformation".

We're making the same mistake with AI. And just like the early web, the companies that figure out how to think differently about AI - not just as a faster tool, but as a fundamentally different way of solving problems - those are the companies that are going to win big.

Part 1: The Human Scale Problem

Today's newsletter came out of my starting to assemble the 12 Days of AI Use Cases, a series on the Trust Insights blog - days 1 and 2 are up, and there are 10 more before the end of the year. When I look at the use cases that are most frequent in the dataset of 13,000 different use cases, most of them have one thing in common:

They're all very human.

By that, I mean the tasks are human scale tasks:

Write blog posts
Write emails
Summarize videos
Summarize meetings
Create images

These are tasks that are typically done by humans. Perhaps not well, not fluently, but they're very human-sized tasks.

What does that mean? Fundamentally, it means that the most common use cases for AI are to replace human-level tasks, to find efficiencies, to reduce costs, expenses, and burdens. That's not necessarily a bad thing by itself - we all have items on our to do lists that we can't wait for machines to take over.

Summarizing my activities for a client at month's end? Yes please, AI. Take that away. I don't want it. I add no value to that task.

Assigning meeting notes and action items from a call? All yours, AI, with my blessing and gratitude. Besides, you remember more and better than I do what I said on a call.

But that's where the problem is with AI right now. Because we still largely don't understand the technology, we're assigning it human scale, human size tasks. We're having it write social media posts or building Python code snippets, things that are impressive for their quality, but not anything that will make someone go, "holy crap, I didn't know machines could do that".

And more dangerously, we're inculcating a mindset that because AI is so fluent at human-scale tasks, perhaps we don't need to employ nearly as many humans. We see this especially in a lot of the marketing materials and thinking from the tech industry itself.

"Do more with less" has been a mantra of management for decades (ever since the Nixon era, when productivity and wages diverged and never synced back up), and AI is just the latest evolution of that.

But we know where that road ends, and it's not a pretty place.

So what does this mean for AI? What are we doing wrong? It comes down to one simple fact: we are having a hard time thinking beyond human scale.

Understandably, since we're human, this isn't a surprise. It's like accusing a cat of thinking cat-scale and not thinking bigger, not thinking outside the box. (For obvious reasons, I prefer my cats to think inside the box, especially if it's the litter box) But humans - and I include myself in the group, however reluctantly - struggle to think beyond human scale.

Yet if we want to unlock the value of AI and really realize its potential, that's what we have to do.

Part 2: Beyond Human Scale

What does it look like to think beyond human scale? Think about the dimensions of the human mind. We are limited by our biology. The neural network in our heads is massively parallel but as computers go, achingly slow compared to our silicon creations. Our memory is incredibly unreliable. Our emotions often get in the way, especially when it comes to our decision-making processes. And our cognition, while incredibly sophisticated, is also limited in terms of the amount of information we can actively process at one time.

All that adds up to limitations we bring to everything we do. To transcend human scale, to unlock the value of AI, we need to extrapolate past our limitations.

What does that look like? It's five major dimensions, five foundational use case categories where AI (and computers in general) dramatically outshine us:

Speed
Scale
Flexibility
Complexity
Patience

In each of these areas, the limitations of flesh and blood impose severe constraints on what we're able to do. It's these five areas where we will find AI's true value.

Speed: Racing Past Human Constraints

AI, because of the scale of computation, is simply faster than us at most of the tasks we put it towards. Anyone who's ever asked ChatGPT to assemble a 500 word blog post knows that it will create a coherent, if uninteresting, blog post with just a short prompt in seconds. This same newsletter that I've been writing to this point has taken me 22 minutes of typing. Compared to a large language model, I am slow as molasses.

When it comes to solving problems with AI and finding valid use cases, the first dimension we should look for is whether our human limitations are imposing a speed limit on our problem. If there is one, we're in the territory of valuable AI solutions.

Scale: Surpassing Biological Memory Limits

The amount of knowledge that the human brain can hold and accurately recall is surprisingly limited. We have different kinds of memory (working, declarative, non-declarative, etc.) but it's clear both from academic study as well as practical experience that our minds and memories are only so reliable, especially when it comes to the management of large amounts of knowledge.

Despite the name "neural networks" and various analogies to human brains, AI in general and generative AI especially work nothing at all like the human brain. When a model is trained, it has a series of statistical associations built within it, mathematical representations of data that it recalls when prompted. That long-term memory, or latent space, is static - models don't learn unless they're retrained - and vast.

For example, Meta's Llama 3 family of models is trained on about 15 trillion tokens of data, or about 10 trillion words. 10 trillion words is a bookshelf of books that stretches around the equator of the planet, twice. No human mind can even come close to memorizing that information at all, much less accurately recall it.

And despite things like hallucinations, AI is surprisingly good at recall without regard to importance. We humans remember things that are important. We don't remember what we had for lunch four weeks ago. If knowledge exists in a model's long-term memory, it's capable of fishing it out mostly correctly.

That scale of memory, recall, and data is far beyond human scale. When it comes to identifying AI use cases, we should be thinking about the scale of data we're working with. If a problem needs to be solved with a scale of data beyond the human mind, it's a candidate for AI.

Flexibility: Adapting Across Boundaries

Quick, translate this sentence into French:

"I'd like to order two golf shirts in extra large size, light blue."

Got it? If you're fluent in French, that was probably easy. If you're not, then you couldn't do it.

Now translate it into Swahili. And German. And Ukrainian. And 11th century English. And Akkadian from 2330 BCE.

There are very few people on the planet who are polyglots in such a wide, wild assortment of languages.

When we ask the same of a generative AI model, we get:

"Je voudrais commander deux polos en taille extra large, bleu clair."
"Ningependa kuagiza mashati mawili ya gofu ya saizi kubwa ya ziada, rangi ya samawati."
"Ich hätte gerne zwei Golfshirts in XL, hellblau."
"Я хотів би замовити два гольф-поло великого розміру, світло-блакитного кольору."
"Ic wōlde biddan twā syrcan, swīðe greates sīðes, blǣw of heofone."
"šitti ṣubātī rabûti uqnî ša šamê lu aknuk"

That level of linguistic fluidity is far beyond human scale - and the models take into account practical limits, like the fact that golf shirts did not exist in Ancient Babylon, nor did golf. (golf wouldn't come around for another 40 centuries)

Equally true, we can ask a language model to compose song lyrics, love poetry, and Google Analytics reports, and the results will all be satisfactory, if not great.

The very architecture of AI is such that if the model has seen examples in its training data, it can faithfully infer from those examples what it's supposed to be doing. And the limitations on this are very few; it doesn't take much to get knowledge into a language model. A recent study showed you only needed a couple hundred samples of anything for a model to be able to process similar data.

Flexibility is a hallmark of AI, far beyond what any individual human is capable of. When it comes to identifying AI use cases, we should be thinking about the necessary flexibility required for a task. The more mental flexibility and fluidity required, the better the use case for AI.

Complexity: Mastering the Intricate Webs Humans Struggle With

Humans are generally good at pattern recognition. We remember faces, we keep track of the seasons, we know the smell of our former relationships' perfume or cologne with just a hint of a scent, we recognize the laughter of our child or the bark of our dog in milliseconds.

But we're easily overwhelmed. Finding a face in a crowded nightclub takes longer than in a coffee shop. Identifying a trend in a sea of data is almost impossible for us (which goes back to scale issues). And seeing tons of weak signals and distilling them down into a strong signal is so rare a talent, people who have it can command astonishing salaries, especially in financial services.

There's a reason why doctors have surprisingly high rates of false diagnosis on life threatening illness - the British Medical Journal found an error rate of 11.1% for human doctors in diagnostic errors of serious conditions with a 4.4% adverse outcome rate (i.e. death or permanent disability), harming a million people a year needlessly.

Part of that is because our pattern recognition breaks down under stress, with fatigue, or when we're overwhelmed with too much data. We lose the ability to recognize the necessary patterns because our brains process information in a particular order and those adverse conditions disrupt our cognition.

AI processes information very, very differently. When you look at how AI analyzes a prompt, for example, it analyzes everything all at once. As a result, it can connect the dots in vast quantities of data, identifying patterns that our minds as linear processors just don't do well. Those complex tasks, complex pattern recognition skills, are where AI really shines - both classical and generative AI.

If we're looking for patterns at scale, we're talking about AI use cases. Humans aren't up to the task beyond the abilities of our working memories, but machines are. The more complex the pattern and the bigger the data, the better the use case for AI.

Patience: Enduring the Tedious Without Error

Finally, that most human of failings: patience, or the lack thereof. Because machines aren't sentient, have no self-awareness, and have no emotions, they can be endlessly patient. Nothing bothers them, nothing bores them, nothing frustrates them - and the opposite is true for humans.

Any task that is highly repetitive creates the conditions for boredom, for carelessness. Hundreds of people die per year in the construction industry, falling off tall buildings. You would think being in a life or death situation would make you extra careful, but repetition breeds that carelessness.

Consider tasks like staffing a call center. You can only answer the same question over and over again so many times before you're bored or irritated. A machine can do that all day long and be just as chipper, polite, and eager on the last call of the day as the first one.

This is where AI truly shines - it is endlessly patient. Consider the needs of a neurodivergent child who needs things repeated to them over and over, or asks the same question in class over and over. A human teacher inevitably gets frustrated with the child. A machine performing the same task will answer the question in the same manner even if the child has asked it 15 times in a row.

When we need patience - and when humans fail to deliver consistently - we are in the realm of AI. The more patience needed, the better the use case for AI.

Part 3: Identifying Use Cases

The challenge with superhuman scaled problem identification is that... well, we're not superhuman. We're all regular humans. And while we can solve existing problems if we've identified them as superhuman problems, we're not necessarily great at identifying novel problems and solutions that could create enormous value, but we don't see how yet. In the Rumsfeld Matrix, we don't know what we don't know.

For example, those of you who have the same color hair as me probably remember the early days of the Web. For the first 10 years or so - 1994 to 2004 - almost all company websites were basically the company's brochure, just available digitally. Only a few companies did really well at converting their legacy operations to the digital era; most of the companies we know from that period are startups that grasped the new paradigm, like the online bookseller Amazon.com.

We're at the same point with AI today. We're tasking AI - a novel solution - with old problems and old approaches. Critics and skeptics of AI often complain that there's no value in AI because it's not solving old problems with old solutions as effectively as humans do, much in the same way I suppose early critics of cars probably complained that cars were less competent and more dangerous than horses.

Where you will find value in AI is looking at the big picture business challenges you have and seeing how those challenges map to the big five criteria. What major challenges, what big problems could be solved by a combination of speed, scale, flexibility, complexity, and patience?

A problem that can be solved by one dimension, like speed, is probably something AI can help with, but you won't see gargantuan changes in your business. Making things faster is good, and there's value in that, but it's not going to be transformative. If you can make things scale and be faster and identify previously inaccessible patterns in your data? Now you're talking about transformative capabilities.

I often say in my keynote talks that using AI to write blog posts is like taking a Harrier military jet to the grocery store to pick up groceries. Yes, it can do that, but it's capable of so much more. The challenge is for us to recognize that a Harrier isn't a faster car. It's a different vehicle entirely. AI isn't a faster intern or a better calculator. It's a different method of computation entirely, and the sooner we wrap our brains around that, the sooner we will get more value out of it.

Here's an example. Suppose you wanted to pitch a podcast. Existing methods and knowledge and practices suggest you hit the podcast's website and fill out the contact form with your pitch, mentioning the podcast by name, alluding to how much you love the show even though you haven't heard a minute of it, and making your pitch.

Applying AI to this particular problem in a human mindset would be to use generative AI to write a slightly more interesting pitch, or perhaps some Python code to fill out website forms faster. That's solving the old problem.

Thinking about speed, scale, complexity, flexibility, and patience, you could take the last 50 episodes of the podcast, transcribe them with speech to text code, then feed that to a generative model to identify some interesting hidden patterns. You'd then have AI generate an audience persona based on those patterns, perhaps even construct a rudimentary inferred SWOT analysis, and then craft a pitch that aligns your point of view not only with the podcast's literal words, but everything you've inferred from an AI's vast knowledge of all podcasting and media.

What you'd send to the podcaster was not just a pitch, but perhaps a valuable analysis of ways that their show could get more audience or retain more listeners... and book you as a guest as well. That's using the superhuman capabilities of AI far in excess of what's feasible for a person to do today. Could a single person listen to the last 50 episodes? Sure. But they probably could not reliably recall everything in all 50 episodes, nor derive the hidden patterns within them, nor come up with an inferred SWOT analysis from it, at least not in a way that wouldn't take weeks of effort.

Just that straightforward example shows you how different a process like podcast pitching would be if you applied the superhuman capabilities of AI and stopped trying to make it do things the old way.

Part 4: Wrapping Up

Here's your next step: Pick your biggest business challenge. Not the day-to-day stuff, but the gnarly problem that keeps you up at night. Now run it through our five dimensions - speed, scale, flexibility, complexity, and patience. If you're hitting at least three of these dimensions, you've got a legitimate AI use case that could transform your business.

The companies crushing it with AI right now aren't the ones writing better blog posts. They're the ones who looked at their most painful problems and asked "What if we didn't have human limitations?" That's your roadmap. That's your opportunity.

Stop trying to make AI a faster horse. Start thinking about what you could do with a rocket ship.

(and if you'd like content like this delivered as a workshop at your company or event, let me know.)

How Was This Issue?

Rate this week's newsletter issue with a single click. Your feedback over time helps me figure out what content to create for you.

Here's The Unsubscribe

It took me a while to find a convenient way to link it up, but here's how to get to the unsubscribe.

If you don't see anything, here's the text link to copy and paste:

https://almosttimely.substack.com/action/disable_email

Share With a Friend or Colleague

If you enjoy this newsletter and want to share it with a friend/colleague, please do. Send this URL to your friend/colleague:

https://www.christopherspenn.com/newsletter

For enrolled subscribers on Substack, there are referral rewards if you refer 100, 200, or 300 other readers. Visit the Leaderboard here.

Advertisement: Bring Me In To Speak At Your Event

Elevate your next conference or corporate retreat with a customized keynote on the practical applications of AI. I deliver fresh insights tailored to your audience's industry and challenges, equipping your attendees with actionable resources and real-world knowledge to navigate the evolving AI landscape.

👉 If this sounds good to you, click/tap here to grab 15 minutes with the team to talk over your event's specific needs.

If you'd like to see more, here are:

ICYMI: In Case You Missed it

Besides the Generative AI for Marketers course I'm relentlessly flogging, this week, I had a BIG series on how to use generative AI to land your next job. The posts and YouTube videos are available, but it's also available as a free course on the Trust Insights Academy - and that version has the prompts you can use. Plus, that's easier to share if you know someone looking for work.

The 12 Days of AI Use Cases Series

Skill Up With Classes

These are just a few of the classes I have available over at the Trust Insights website that you can take.

Premium

Free

Advertisement: Generative AI Workshops & Courses

Imagine a world where your marketing strategies are supercharged by the most cutting-edge technology available – Generative AI. Generative AI has the potential to save you incredible amounts of time and money, and you have the opportunity to be at the forefront. Get up to speed on using generative AI in your business in a thoughtful way with Trust Insights' new offering, Generative AI for Marketers, which comes in two flavors, workshops and a course.

Workshops: Offer the Generative AI for Marketers half and full day workshops at your company. These hands-on sessions are packed with exercises, resources and practical tips that you can implement immediately.

👉 Click/tap here to book a workshop

Course: We’ve turned our most popular full-day workshop into a self-paced course. Use discount code ALMOSTTIMELY for $50 off the course tuition.

👉 Click/tap here to pre-register for the course

If you work at a company or organization that wants to do bulk licensing, let me know!

Get Back to Work

Folks who post jobs in the free Analytics for Marketers Slack community may have those jobs shared here, too. If you're looking for work, check out these recent open positions, and check out the Slack group for the comprehensive list.

Advertisement: Free Generative AI Cheat Sheets

Grab the Trust Insights cheat sheet bundle with the RACE Prompt Engineering framework, the PARE prompt refinement framework, and the TRIPS AI task identification framework AND worksheet, all in one convenient bundle, the generative AI power pack!

Download the bundle now for free!

How to Stay in Touch

Let's make sure we're connected in the places it suits you best. Here's where you can find different content:

My blog - daily videos, blog posts, and podcast episodes
My YouTube channel - daily videos, conference talks, and all things video
My company, Trust Insights - marketing analytics help
My podcast, Marketing over Coffee - weekly episodes of what's worth noting in marketing
My second podcast, In-Ear Insights - the Trust Insights weekly podcast focused on data and analytics
On Threads - random personal stuff and chaos
On LinkedIn - daily videos and news
On Instagram - personal photos and travels
My free Slack discussion forum, Analytics for Marketers - open conversations about marketing and analytics

Listen to my theme song as a new single:

Advertisement: Ukraine 🇺🇦 Humanitarian Fund

The war to free Ukraine continues. If you'd like to support humanitarian efforts in Ukraine, the Ukrainian government has set up a special portal, United24, to help make contributing easy. The effort to free Ukraine from Russia's illegal invasion needs your ongoing support.

👉 Donate today to the Ukraine Humanitarian Relief Fund »

Events I'll Be At

Here are the public events where I'm speaking and attending. Say hi if you're at an event also:

Social Media Marketing World, San Diego, April 2025
Content Jam, Chicago, April 2025
SMPS, Columbus, August 2025

There are also private events that aren't open to the public.

If you're an event organizer, let me help your event shine. Visit my speaking page for more details.

Can't be at an event? Stop by my private Slack group instead, Analytics for Marketers.

Required Disclosures

Events with links have purchased sponsorships in this newsletter and as a result, I receive direct financial compensation for promoting them.

Advertisements in this newsletter have paid to be promoted, and as a result, I receive direct financial compensation for promoting them.

My company, Trust Insights, maintains business partnerships with companies including, but not limited to, IBM, Cisco Systems, Amazon, Talkwalker, MarketingProfs, MarketMuse, Agorapulse, Hubspot, Informa, Demandbase, The Marketing AI Institute, and others. While links shared from partners are not explicit endorsements, nor do they directly financially benefit Trust Insights, a commercial relationship exists for which Trust Insights may receive indirect financial benefit, and thus I may receive indirect financial benefit from them as well.

Thank You

Thanks for subscribing and reading this far. I appreciate it. As always, thank you for your support, your attention, and your kindness.

See you next week,

Christopher S. Penn